Google CEO: Learn AI Agents Now Or Spend 2027 Playing Catch-Up

So, Neil, I saw— did I ever show you this chart? I mean, we like to all talk about LVMH, and I thought this was really interesting. So LVMH, obviously they own Louis Vuitton, they own Dior, they own Fendi, they own Sephora and 70+ luxury houses, right? So this chart over here I thought was fascinating. So if you look at since the highs of— and Neil, you can read the charts over here, but you see December 2021, this is obviously during the height of the pandemic.

Yeah.

And so everyone's flush with cash, right? And so, you know, you saw the lines around like the Beverly Hills Louis Vuitton over here. So December '21, 44% revenue growth, right? And then you move to March of '22, it drops to 37%. June 2022, 32%. So this chart is just, it's just, it's like a hill that's going downwards, right? And you can see it goes all the way to December 2025. And it goes negative 5%, right? And then it just keeps dropping, dropping, dropping. So LVMH has now put up 16 consecutive quarters of decelerating growth. And I think they've done a great job with marketing overall with this stuff. But what do you think is happening here?

So I don't really look at it as 16 quarters of decelerating growth. I know the numbers are showing that, But here's what people don't talk about. It is a company that's been around for a long time. Historically, you know, as they've scaled, because they typically buy brands, even though they're a fashion house, they're more like private equity in which they're just buying other people's fashion brands, tucking them in, running their playbook, scaling up, and just holding them for a very long time. When you look at LVMH in 2021, I'm actually looking at their stock chart right now. They grew exactly 43.82% in revenue. That's a massive growth in one year, especially for a company at that time in 2021, they did $64 billion. In 2023, which was their peak, they did $86.15 billion. 2024, it went down 1.71% to $84.68 billion. If you look at 2025, it went down to $80.81 billion, and they're still shrinking a little bit here and there. But the way I look at LVMH is if you standardize the curve and you take away the COVID bump, they've actually grown really nicely compared to where they started pre-COVID. And yes, people had a lot more discretionary spending during COVID because the government deposited money in people's bank accounts and a lot of people ended up using it on things like luxury goods.

So here's what I'll call out. Remember, we're talking about revenue growth here just so everyone sees this, right? So you had 44% revenue growth and now It's negative 5%, right? And so, but you have EBITDA margins held in the mid-30s, right? And then it went back, it dropped back a little bit. It's still a great, I wouldn't even say it's a business, it's a collection of businesses and they're still very well known anywhere that you go in the world. So I still think it's a great business. I mean, I would love to own it. So I think Chamath actually posted this just saying, it's, you know, is it over, right? And I don't think it's over. I think, you know, they're going to do just fine. And I think if anything, these brands have been around for quite a long time too. So it's the staying power that the name behind these is— what should we call it? It's brand power. That's what it is. When you think about the 7 powers, brand power fits in for all these different brands that I just mentioned, whether it be Fendi, Dior, Louis Vuitton, and brands like that.

Let's take Louis Vuitton for example. It was founded in 1854. It's a 172-year-old brand. The value in these companies isn't the purse, isn't the handbag. It really is the brand. People buy their products because of that logo. And when you look at Louis Vuitton, there's people who sell knockoffs, similar lookalikes under different brands. And even if they're 1/20th of the price, people still don't want them because they're not LV. People take pride in owning an LV handbag or purse or whatever they're called, or their clothes or their sneakers. Like, their brands have staying power. I believe they also own Tiffany's as well. I could be wrong though. Uh, is it Tiffany's owner? Yes, LVMH owns Tiffany as well, the diamond company.

Yep. So anyway, I thought that'd be a great place to start, but I want to show you one more thing here. So this is a Sundar Pichai talking about this, right? So CEO of Google. And so he says this, if you don't learn how to orchestrate agents now, you'll spend 2027 catching up to people who started today. So in 30 minutes, he explains why the best engineers stopped writing code and started running agents, right? So right now you have this craze right now on X where the creator of OpenClaw, the creator of Claude Code, I think even Jensen Huang was saying this, Andrew Andrew Ng, who's like a computer scientist as well, well-known computer scientist. They're all talking about building loops, right? These are the next evolution where you don't prompt anymore, you do loops, and then you just have these loops run over and over, and you just have these agents just continue to improve over time, right? And it's on its way to recursive self-improvement. And so going back to Sundar's quote here, if you don't learn how to orchestrate agents today or now, you'll spend 2027 catching up to people who started today. I just think there's very much a red herring, and I continue to see this when we're talking to people, when we're interviewing people, right? Like yesterday, I had talked to a handful of people, and it's what we do now to kind of see if people have it or they don't is we just get them to open up, hey, like, oh, what's the coolest workflow that you built? Okay, cool, great. Can you open up your Codex? Can you open up your Claude Code? We see their usage, we see how they're using it, we get them to talk through it, but I do believe That at least for the next few years, maybe the long term, maybe it just smooths out. If you don't learn this today, what's going to happen is, let's say Neil learns this today and then I don't learn it. By next year, if I'm trying to catch up to him, he's already off in another planet, right? Because this stuff is compounding, it's changing every single day. So I think it's interesting that he's saying that. And because my thing, I think I've asked you this, Neil, we've kind of talked about this a little bit. I was just like, man, if You start compounding today, you're 12 months ahead of me, and I start compounding after that, how am I supposed to catch you? And but then like my logic was, well, the internet, like everyone was still kind of evened out at the end of the day. So who knows how it'll play out because we've never really had synthetic intelligence like this. So I thought that was just an interesting quote.

Yeah, when you look at the compounding, I do agree there's going to be some sort of compounding effect, but what is the net result from that compounding effect. And I'll give you a prime example of this is you look at Uber. Uber talked about how they went through their tokens. I think it was in a matter of months, and this was tokens that they had allocated for a whole year. But if you look at Uber's stock price, all the extra usage, all the cranking out of features hasn't really moved their revenue and caused a lot more revenue growth than one may think, right? So just because you do 10x more the volume or 20x more the output or even 1,000x, it doesn't mean that translates into 1,000x or even 100x or even 2x more revenue. So are you going to have more work to catch up on? Yes. Does all that work that you have to catch up on matter? It's yet to be seen. And I think a lot of it won't produce revenue. We're seeing the same thing in marketing. There's so many people that if you push AI aside, whether they have a bigger team or whether they have more software or more technology or leveraging AI, there's some teams that do 10 times more marketing than others. And it's not always the team that's doing 10 times more that's winning. Sometimes the team that's doing 1/10 the stuff is just focusing on 1 or 2 channels that's really driving a lot of results and they fine-tuned it and they're so good at it and they figured out how to really lock other people out from Getting into that channel. And in those cases, you know, you don't need to do 10x this stuff. That one channel, assuming you can keep milking it and you, you know, can figure out ways to making it sustained, could potentially cause you to be a much larger business.

Real quick, if you want to acquire customers faster and more efficiently this year with the latest strategies and tactics, then check out singlegrain.com. That is my ad agency. Again, www.singlegrain.com. Check it out. And if it seems like a fit, we'll get in touch and help you with a free marketing plan. So as it relates to that, I'm actually going to pull up this tweet from Glenn Gabe. So Google here, speaking of AI for a second, just continuing on the Google piece. Like I've said before, there's no way Google lets companies spamming for AI search continue. They will ramp up systems and add manual actions. Google research shows how AI spam can be detected. So basically they're adding in these content defenders, right? Which I'm just like, wait, Wait, isn't the algorithm already the defender? But so now they're using something, um, I don't even know what SBERT is, maybe SearchBERT, for identifying AI-generated content. So this is what Google is doing now, right? Because keep in mind, a lot of people have been spamming these best-of listicles and listing them at the top, and then these LLMs are picking them up as well. So what will probably be the most of interest is that researchers acknowledge that the use of SentenceBERT as a way to identify semantically similar sentences. They cite SentenceBERT to validate a core assumption of their paper, that automated AI-generated text leaves a distinct mathematical footprint, aka text embeddings, that can be detected. So this system is said to be a highly accurate defense against coordinated generative AI spam, which means something like this could conceivably be in use. The new system is called Scalable Cluster Termination System. Oh my God, so many names. SCTS. And the research paper paper, Scalable Detection of Adversarial Synthetic Slop and Coordinated Media Abuse.

Dude.

Okay. Anyway, so all that to say is that what we've called out even a few months ago and probably even a few years ago is that you're better off doing the black hat stuff in a white hat way. So if you're going the extra mile with your— you have good data-backed content, you're doing good data storytelling, right? That's going to go a lot further than you just stealing someone else's graph and just reposting.

Yeah.

So I think Google's going to be fine. I think Claude's going to copy this. I think OpenAI is going to copy this. I think Google's going to have the best— like, I will say this, Neil, like, how often do you use AI mode right now?

Not that much, but I don't need to use AI mode because when I ask Google a specific question, right, and I'll show an example on my phone. Here, like, um, if I put fastest in the world. So I'm actually going to close these windows out because it's not working on my phone. What's a random question? Um, here, I'll type in something right now. I'm sick and I have a sore throat.

Oh yeah.

Fever.

Neil went on a cruise, guys. What do you— This is why he's low energy Neil today. LEN, low energy Neil. There's another acronym for you.

Okay, so as you can see here, I typed it in, some misspellings. I clicked show more in the AI overviews area, right? And it's telling me what I think it thinks I have. And then I can ask any follow-up question if I scroll to the bottom. Type and answer. So whether you want to call that AI mode or AI overviews, they have it right then and there in the AI overviews where you can just interact with it, kind of like ChatGPT. And it's simple. I don't have to go to a specific URL. I just type and it works.

Yeah. So I use it. I mean, I would say I use it a few times a week and it's been pretty solid in my opinion. And so I think Google's always going to have at least for the next couple years, the best search experience because they have all these defenders and they've been fighting spam for such a long time. So speaking of AI search, by the way, there's a good— Eli Schwartz tweeted this, so check this out. So Eli Schwartz is a strong SEO, wrote a book called Product-Led SEO. So basically he's saying— well, was that a sneeze, Neil? What was that?

Cough.

Okay, so if you want to win in AI search, it comes down to 3 things. You got to be unique, you got to be helpful, and be agent ready. So the first 2, by the way, this is all— this has always been Google's rules, right? Be unique and be helpful with the content, but also be agent ready. So he's not talking about secret schema or chunking tricks or file uploads that make LLMs fall in love with their site. He's not talking about that, but it's 3 plain words, right? So this should worry every AEO tool or consultant currently charging 5 figures a month for a checklist. So the 3 pillars. Okay, so it says 3 plain words. I'm looking for the 3 plain words over here. Okay, so almost every tactic built around gaming AI has a shelf life. Okay, we've talked about that before because the moment enough people exploit it, Google quietly patches it, which is what you've just seen with us talking about Expert and all this other stuff, these defenders. Okay, structured data formatting work. I mean, that to me is just standard optimization. And then this exposes something funnier, the entire category of tools that will happily tell you your brand's visibility score across ChatGPT, DeepSeek, AI overviews, color-coded and exportable for a deck for your CMO. Okay. Okay. All right. Where's the 3 words? This is a problem with some of these posts. They just take so long to get to the point. Agent-led spam doesn't count if it doesn't move the needle. The 3 pillars Google name doesn't have that for expiration problem. Okay. What?

I don't know where the 3 words are.

Dude, Eli. What the hell, man? Okay. The smartest move right now isn't chasing whether new AEO trend is circulating on LinkedIn this particular week, and it does involve refreshing a tracking dashboard, hoping the number moves on its own. Okay. So user empathy and marketing. Okay. So again, okay, it's right here. Here it is, Neil. Here's the payoff. Be unique, be helpful, and be agent ready. Those are the 3 words. That's all you need to do to win an AEO. Eli, I hate you for writing this post. It could have just been this graphic of this.

I don't mind the post. I like it. But you know, you said it earlier, leveraging black hat tactics and white hat ways. Example that we all know right now is they're leveraging listicles like crazy to figure out what to cite. Both Google is, ChatGPT is, Perplexity, pretty much all of them are. Now what a lot of people are doing is publishing these listicles on their own websites and they're hoping that it gets ranking. What some people are doing is just spamming guest posts on other people's websites and using AI to write listicle content that places them number 1. So if you want to talk about a black hat strategy in a white hat way, if you do something unique in your company, so you could have free products, free services, you know, you go above and beyond, maybe you share some data or you create some data, pay for some research, you do something unique that makes you stand out. And then you go and build relationships with journalists. Let's say if it's Forbes or entrepreneur.com or TechCrunch or whatever it may be, right? Journalists within your space. And then you reach out to them and try to get them to include you in a listicle or talk about agencies. Let's say you won an award like we won Performance Marketing Agency of the Year from Ad Age, but we pitch journalists, try to get them to talk about it. And talk about why we won, who we came— who we went up against, and why we were chosen and some of the results. It's not a pure listicle, but when they're talking about how you won and who you went up against, it kind of is like a listicle, just another version of it, not as explicit. And when they talk about what you did, this is creating positive sentiment, right? Because they're breaking down what you did to win the award and the results that you provided, so it's making your agency look better than others. Now, Forbes may write on it, most cases they won't, but if you do this with enough of them, you're going to get some of them to write on it and it helps you out.

That's it for today and we will see you tomorrow.